Exploiting gene deletion fitness effects in yeast to understand the modular architecture of protein complexes under different growth conditions

Background Understanding how individual genes contribute towards the fitness of an organism is a fundamental problem in biology. Although recent genome-wide screens have generated abundant data on quantitative fitness for single gene knockouts, very few studies have systematically integrated other types of biological information to understand how and why deletion of specific genes give rise to a particular fitness effect. In this study, we combine quantitative fitness data for single gene knock-outs in yeast with large-scale interaction discovery experiments to understand the effect of gene deletion on the modular architecture of protein complexes, under different growth conditions. Results Our analysis reveals that genes in complexes show more severe fitness effects upon deletion than other genes but, in contrast to what has been observed in binary protein-protein interaction networks, we find that this is not related to the number of complexes in which they are present. We also find that, in general, the core and attachment components of protein complexes are equally important for the complex machinery to function. However, when quantifying the importance of core and attachments in single complex variations, or isoforms, we observe that this global trend originates from either the core or the attachment components being more important for strain fitness, both being equally important or both being dispensable. Finally, our study reveals that different isoforms of a complex can exhibit distinct fitness patterns across growth conditions. Conclusion This study presents a powerful approach to unveil the molecular basis for various complex phenotypic profiles observed in gene deletion experiments. It also highlights some interesting cases of potential functional compensation between protein paralogues and suggests a new piece to fit into the histone-code puzzle.


Table S2: Fraction and enrichment of unknown genes in complexes in the strong negative effect category
Fraction and enrichment of genes in complexes compared to genes not in complexes in the strong negative fitness effect category, considering only those genes annotated as 'Uncharacterized ORF'. (i) essential and inessential genes, (ii) only inessential genes. P-values were computed using a one-sided Fisher's exact test. Enrichments are given on a log 2 -scale. Fermentable:

Figure S1: Comparison of the fitness of yeast strains upon deletion of genes in complexes and those not in complexes
Distributions of strain fitness upon deletion of genes in complexes (red) and genes not part of complexes (blue) in the two fermentable and the three nonfermentable media considered. Genes with a fitness of zero are essential. The fitness values of individual genes are partitioned into four categories: 'strong negative effect' (--), 'moderate negative effect' (-), 'weak or no effect' (0) and 'positive effect' (+). Different shades of red illustrate the percentage of genes in complexes (for which we have essentiality data) in the four fitness categories, with deep red corresponding to 100% (1404 genes). Different shades of blue illustrate the percentage of genes not in complexes (for which we have essentiality data) in the four fitness categories, with deep blue corresponding to 100% (3770 genes). Enrichments are given on a log 2 -scale. Fermentable: Non−fermentable:

Figure S2: Comparison of the fitness of yeast strains upon deletion of inessential genes in complexes and those not in complexes
Distributions of strain fitness upon deletion of inessential genes in complexes (red) and inessential genes not part of complexes (blue) in the two fermentable and the three non-fermentable media considered. The fitness values of individual genes are partitioned into four categories: 'strong negative effect' (--), 'moderate negative effect' (-), 'weak or no effect' (0) and 'positive effect' (+). Different shades of red illustrate the percentage of genes in complexes (for which we have essentiality data) in the four fitness categories, with deep red corresponding to 100% (868 genes). Different shades of blue illustrate the percentage of genes not in complexes (for which we have essentiality data) in the four fitness categories, with deep blue corresponding to 100% (3350 genes). Enrichments are given on a log 2 -scale. Fermentable: Non−fermentable:

Figure S3: Comparison of the fitness of yeast strains upon deletion of genes in MIPS complexes and those not in MIPS complexes
Distributions of strain fitness upon deletion of genes in the hand-curated set of 266 yeast complexes in the Munich Information Center for Protein Sequences (MIPS) database [1] (red) and genes not part of MIPS complexes (blue) in the two fermentable and the three non-fermentable media considered. Genes with a fitness of zero are essential. The fitness values of individual genes are partitioned into four categories: 'strong negative effect' (--), 'moderate negative effect' (-), 'weak or no effect' (0) and 'positive effect' (+). Different shades of red illustrate the percentage of genes in complexes (for which we have essentiality data) in the four fitness categories, with deep red corresponding to 100% (1134 genes). Different shades of blue illustrate the percentage of genes not in complexes (for which we have essentiality data) in the four fitness categories, with deep blue corresponding to 100% (4040 genes). Enrichments are given on a log 2 -scale. Fermentable: Non−fermentable:  Fermentable: Non−fermentable:

Figure S5: Comparison of the fitness of yeast strains upon deletion of genes in Krogan complexes and those not in Krogan complexes
Distributions of strain fitness upon deletion of genes in the set of 547 yeast complexes defined by Krogan et al. [2] (red) and genes not part of Krogan complexes (blue) in the two fermentable and the three non-fermentable media considered. Genes with a fitness of zero are essential. The fitness values of individual genes are partitioned into four categories: 'strong negative effect' (--), 'moderate negative effect' (-), 'weak or no effect' (0) and 'positive effect' (+). Different shades of red illustrate the percentage of genes in complexes (for which we have essentiality data) in the four fitness categories, with deep red corresponding to 100% (2515 genes). Different shades of blue illustrate the percentage of genes not in complexes (for which we have essentiality data) in the four fitness categories, with deep blue corresponding to 100% (2659 genes). Enrichments are given on a log 2 -scale. Non−fermentable:

Figure S6: Comparison of the fitness of yeast strains upon deletion of inessential genes in Krogan complexes and those not in Krogan complexes
Distributions of strain fitness upon deletion of inessential genes in the set of 547 yeast complexes defined by Krogan et al. [2] (red) and inessential genes not part of Krogan complexes (blue) in the two fermentable and the three non-fermentable media considered. The fitness values of individual genes are partitioned into four categories: 'strong negative effect' (--), 'moderate negative effect' (-), 'weak or no effect' (0) and 'positive effect' (+). Different shades of red illustrate the percentage of genes in complexes (for which we have essentiality data) in the four fitness categories, with deep red corresponding to 100% (1871 genes). Different shades of blue illustrate the percentage of genes not in complexes (for which we have essentiality data) in the four fitness categories, with deep blue corresponding to 100% (2347 genes). Enrichments are given on a log 2 -scale.

of complexes in which a gene is present and the number of potential interactors
Box-and-whisker plots of the number of potential interactors of genes present in multiple complexes, based on the matrix model [3] which assumes that each protein interacts with each other protein in a complex. Start and end of the boxes indicate the first and third quartile of the number of interactors distribution of genes present in a given number of complexes, and whiskers denote the respective minimum and maximum number of interactors. The medians of the respective distributions are shown as black bars. As only 21 genes are present in more than 16 complexes, we grouped them together. The gamma correlation coefficient between the number of complexes in which a gene is present and its number of potential interactors is 0.83.

yeast strains upon deletion of inessential genes present in multiple complexes
Box-and-whisker plots of strain fitness upon deletion of inessential genes which are part of multiple complexes, measured in the two fermentable and the three non-fermentable media considered. Start and end of the boxes indicate the first and third quartile of the fitness distribution of inessential genes present in a given number of complexes, and whiskers denote the respective minimum and maximum fitness values. The medians of the respective distributions are shown as black bars. As only 21 inessential genes are present in more than 16 complexes, we grouped them together.  Non−fermentable: Figure S10: Comparison of the fitness of yeast strains upon deletion of genes unique to complex cores and genes unique to attachments Distributions of strain fitness upon deletion of genes only present in cores (red) and genes only present in attachments (blue) in the two fermentable and the three non-fermentable media considered. Genes with a fitness of zero are essential. The fitness values of individual genes are partitioned into four categories: 'strong negative effect' (--), 'moderate negative effect' (-), 'weak or no effect' (0) and 'positive effect' (+). Different shades of red illustrate the percentage of genes only present in cores (for which we have essentiality data) in the four fitness categories, with deep red corresponding to 100% (337 genes). Different shades of blue illustrate the percentage of genes only present in attachments (for which we have essentiality data) in the four fitness categories, with deep blue corresponding to 100% (320 genes). Enrichments are given on a log 2 -scale.